Chips/parts you hate working with...

[David Hess]

I have avoided Atmel products in the past because of their poor datasheets.  There are examples of even worse datasheets like from Allegro but that does not make Atmel's datasheets any better.  TI's datasheets have gotten significantly worse over time but I have other reasons to avoid their parts.

[T3sl4co1l]

Atmel's documentation is usually clear, well formatted, sufficiently redundant (similar sections are often near copy-paste -- saves on time having to read through different sections), and well enough explain the code or registers or logic or whatever.  They do lack in the implementation / detailed / analog department though.  Example: the crappy ADC, BOD and POR in the ATmega series, and good luck finding any guaranteed min/max parameters on them (even a lot of typ are missing).

Also the same sorts of things from Altera: good luck finding I/O pin characteristics, power supply specs (bypassing, current consumption..), that sort of thing.

Tim

[Alex30]

Atmel's documentation is usually clear, well formatted, sufficiently redundant (similar sections are often near copy-paste -- saves on time having to read through different sections), and well enough explain the code or registers or logic or whatever.  They do lack in the implementation / detailed / analog department though.  Example: the crappy ADC, BOD and POR in the ATmega series, and good luck finding any guaranteed min/max parameters on them (even a lot of typ are missing).

Also the same sorts of things from Altera: good luck finding I/O pin characteristics, power supply specs (bypassing, current consumption..), that sort of thing.

Tim

Yeah I agree with this. To be honest I suppose it depends on what your expectations are. Any consensus on what microcontroller manufacturer has more detailed datasheets than atmel?

[T3sl4co1l]

Not something I've researched thoroughly at all, but I've looked at an MSP430 or two and they seem to document things well enough.  They're more powerful (16 bit, and usually faster and more features), and very popular, especially for low current consumption purposes (switchable low power clock options, low operating current in general).

Anyone else?

Tim

[suicidaleggroll]

Not something I've researched thoroughly at all, but I've looked at an MSP430 or two and they seem to document things well enough.  They're more powerful (16 bit, and usually faster and more features), and very popular, especially for low current consumption purposes (switchable low power clock options, low operating current in general).

Anyone else?

Tim

I've used several MSP430s in different projects, and have never found anything significant lacking in TI's documentation.  I find them quite pleasant to work with actually, provided they fit the requirements of the application.  I have found a few small bugs in GRACE and of course I hate CCS with a passion and IAR isn't too far behind, but mspgcc works well.

[made2hack]

BNC connectors. I loathe and love them at the same time.

I love the fact that a well shielded SDI cable with BNC connectors will pass data at multi-gigabit speed. And I love the fact you twist to close it.

I also loathe the fact you twist to close it, and that innevitably the connector on the 10000+ euro broadcast equipment feels incredibly flimsy. And that the slightest pull on the cable (because once you've connecte a bnc cable it can never be disconnected again by twisting... I'm looking at you Sony broadcasting equipment) feels as if the entre socket will pull out.

[David Hess]

BNC connectors. I loathe and love them at the same time.

I love the fact that a well shielded SDI cable with BNC connectors will pass data at multi-gigabit speed. And I love the fact you twist to close it.

I also loathe the fact you twist to close it, and that innevitably the connector on the 10000+ euro broadcast equipment feels incredibly flimsy. And that the slightest pull on the cable (because once you've connecte a bnc cable it can never be disconnected again by twisting... I'm looking at you Sony broadcasting equipment) feels as if the entre socket will pull out.

TNC connectors have the solid feel that BNC connectors lack and are easier to use than SMA connectors in test equipment.  Since TNC connectors are basically deprecated, I have grown to like SMB connectors in place of SMA connectors.

One of the pleasant surprises I had when first working with GR-874 connectors is how solid they felt.

[Pedram]

fake/copy chips... i hate theme.

[asgard]

I see, this is an intelligence test, one which those that admit to repeated application of duff parts fail.

We don't always get the choice. I dislike using anything that comes in a QFN or similar package, because even when manufacturers swear they can handle them with no problems at all, I just know I'll end up looking for open circuits where they've not soldered properly. Not ideal when you're trying to find the *real* bugs on a first prototype.

I would agree wholeheartedly.  QFN/DFN parts ( I have been using these for power MOSFET H-bridge drivers, lately) are horrible, when a hobbyist or small-timer like myself doesn't have an X-ray machine to check solder joints.  Another horrible part is the SMD HDMI type A connectors I have been forced to use (from Molex).  This thing puts all 19 pins in a single row adjacent to the main shell.  The clearance is (almost) 0.5mm pitch, and therefore allows at max 6-8mil geometry to run the signals out.  Try to hand assemble a board with one of these.  Half the time the traces just evaporate before the solder joints are solidified.  Next time I am going to try to get connectors from Adam, which just announced double offset rows of through pins on their connector.

[T3sl4co1l]

Half the time the traces just evaporate before the solder joints are solidified.  Next time I am going to try to get connectors from Adam, which just announced double offset rows of through pins on their connector.

Solder iron about... 200C too hot???

Tim

[Gribo]

Certain MGA series LNAs from Avago. Those are so sensitive to ESD, that a single wrong touch can fry them.

[mrkev]

There is nothing worse than fake parts from otherwise (at least we thought) genuine supplier.
I've got parts for a school project (power fets) and they were OK, but the next batch were all fakes. Apparently some half-wit in their management thought that they will make more money if they buy from some chinese reseller and didn't even bothered to check if it meets the spec (and it didn't by far)...

[asgard]

Half the time the traces just evaporate before the solder joints are solidified.  Next time I am going to try to get connectors from Adam, which just announced double offset rows of through pins on their connector.

Solder iron about... 200C too hot???

Tim

SMD parts with hot air at 260C and at least 3" distance from the joint to the air tool tip.  Effective temp at the copper is about 160C.
The Adam parts will allow me to use 10mil traces from the pins, which should improve my disposition something wonderful.

[poorchava]

LCD drivers from Princeton Tech. The serial interface is a bad an sad joke. Total pain in the ass to write software for if that software is to be even remotely efficient. Same thing goed for other chinese products that have derial interfaces.

[mikerj]

You should read more carefully. I wrote IF an Atmel datasheet says the lowest voltage is 3.3V then you shouldn't run the device at 3.3V or you will get burned sooner or later. It is an example of what you can encounter. Other manufacturers like NXP or TI are way more honest about their devices.

Which parts are you referring to specifically?

[Circlotron]

I hate 27K resistors.
Not a technical thing; more like a phobia :-P

[nctnico]

You should read more carefully. I wrote IF an Atmel datasheet says the lowest voltage is 3.3V then you shouldn't run the device at 3.3V or you will get burned sooner or later. It is an example of what you can encounter. Other manufacturers like NXP or TI are way more honest about their devices.

Which parts are you referring to specifically?

All digital chips from Atmel.

[gxti]

You should read more carefully. I wrote IF an Atmel datasheet says the lowest voltage is 3.3V then you shouldn't run the device at 3.3V or you will get burned sooner or later. It is an example of what you can encounter. Other manufacturers like NXP or TI are way more honest about their devices.

Which parts are you referring to specifically?

All digital chips from Atmel.

Many atmega parts go down to 2.7V or even lower. Be more specific.

[nctnico]

Read more carefully what I wrote before: the 3.3V is just an example. If Atmel says a device will run at X volt then add approx 0.2V/0.3V to X to know at which voltage a device actually works reliable.

[T3sl4co1l]

I hate 27K resistors.
Not a technical thing; more like a phobia :-P

I have a better one: 15 ohm resistors.

Don't get it?  Consider this... What color is a 1M resistor?

Tim

[David Hess]

Personally I hate working with any of the PIC 12, 16 or 18 series. The datasheets are fine and they are cheap, reliable parts, but the IDE sucks and their architecture is horrible. These days there is little reason not to use a 24 series though.

The PIC instruction set architecture always reminds me of the 6502.  A little bit of macro assembler messaging can improve the situation for those not comfortable with accumulator based ISAs.

I might prefer Atmel's 8-bit ARV except for the previously mentioned problems with their documentation.  I also have concerns about ongoing availability of Atmel parts.

[coppice]

The PIC instruction set architecture always reminds me of the 6502.  A little bit of macro assembler messaging can improve the situation for those not comfortable with accumulator based ISAs.

Why do you think a PIC is like a 6502? The PIC is Harvard like. The 6502 has a single address space architecture. I would have thought that gave them a very different feel.

[djococaud]

Not high temperature wire you have to solder to a board...with insulation melting and retracting before you try to solder it ! 

[David Hess]

The PIC instruction set architecture always reminds me of the 6502.  A little bit of macro assembler messaging can improve the situation for those not comfortable with accumulator based ISAs.

Why do you think a PIC is like a 6502? The PIC is Harvard like. The 6502 has a single address space architecture. I would have thought that gave them a very different feel.

The separate address spaces hardly make a difference in most microcontroller applications where firmware will be fixed.  The PIC still supports immediate operands even though immediate data gets pulled from the program address space instead of the unified address space.

The programming environment for a processor with unified address space treats program address space and data address space separately most of the time anyway.

[dannyf]

What parts always end up biting you in the ass.....

Early Luminary parts fall into that category. Thank God that TI bought them out.